The writing of timetables, how do they do that?

[whole diary]

[Not from Brighton]

A great many topics on this forum concern the impacts of timetable changes. For example:
http://www.firstgreatwestern.info/coffeeshop/index.php?topic=11481.0

This got me wondering how one goes about changing the timetable, and I rapidly came to the conclusion that it is a very hard problem. Is this indeed the case or am I just a simpleton? To be fair, being as I am a resident of Worcester I initially looked at this city. I wonder if Worcester is especially nightmarish and other, more self contained areas might be easier to analyse?

[grahame]

The more I learn, the more admiration I have for the people who write the timetables, and the people who look after their operation if things aren't running quite to plan.

I have been involved in some "wouldn't it be nice if xxx" scenarios ... and here are the considerations that have come up; I'm sure that I have missed things off this list too.

When is the service wanted by passengers, and from where to where?
Which opposing service will they use for their daily / weekly returns?
Does the service meet the DfT^» (Department for Transport - about) / Franchise specification?
Does the service meet the specification of any other funders?
Will it make money and be good publicity for the TOC^▸ (Train Operating Company)?
Can we do it clockface?

Where does the train come from?
Where does the train go to?
Where does the crew come from?
Where does the crew go to?
Will this fit within shift length and rest break rules?
Will the crew know the road?
Is the available train long enough?
Is the available train a type that will fit on the line's gauge?
Is the available train short enough?
Is the available train appropriate for loading / unloading along the way?
Does the available train have / not have first class as required
Does the available train have have catering as required
Does the available train have enough loo capacity
Does the available train have enough fuel capacity
Will the available train reach a cleaning point often enough?

Where and how well does it need to connect with other trains?
Where and how well does it connect with bus services?
Are the right facilities available where people will be changing trains?
Which platform(s) are available for it?
When can it have the single line?
Can it get through without holding up following trains?
Will it get held up behind trains in front of it?
Are the signalling headways sufficient?
Will it create conflicting movements over junctions?
Are sidings / stabling tracks available when it's not in use?
Can the stations and facilities there cope with the extra traffic generated?
What happens if it gets held up?
What happens if other trains in the timetable get held up?
What happens if a crew member is not available?
What happens if the train is not available?
What happens if the train is unsuitable for public use?
What happens if the train becomes a non-runner?
What happens if a train of a different type is all that's available?
What happens during engineering works?

How long does each leg of the journey take?
How long is required in each station?
How long is required to turn around at the end of the route?
How much recovery / extra time should we add for robustness?
Where do we stop along the way?
How will it be presented in the public timetable?
How long is needed for request stops / how many will actually be used?

Will the customers like and use this service?
Will it be too close to other existing services to be useful?
Will it involve long distances of empty / very quiet running?
Are the fares such that it's economically justified?

And remember that you're looking at the intermeshing of the network, across multiple operators and lines.   It's like throwing a stone into a pond ... it enters the water at one point and ripples spread right out.

[eightf48544]

Whilst you have listed the many issues to be taken into account what must be remembered is that the answers to many of these questions have been accumulated over time and, therefore, as can be seen from the December timetable the bulk of the services are unchanged particularly in the Thames Valley with the major changes, which have been highlighted here, being in the late evening to make it a two track railway ready for electrification and  resignalling.

You are right though that major recasts of the timetable are major undertakings that may appear to work on paper (sub computer nowadays) but require extensive tweaking to get to run smoothly.

Computerisation has helped one of my critisms of BR^▸ (British Rail(ways)) in the Beeching era was that in many parts of the country they still were running pre 1922 timetables and service patterns, if it started from GWR^▸ (Great Western Railway) station it terminated at GWR. Very little rationalisation of routes ocurred notably only at Leeds where the rationalisation took so long and appeared to cost so much that further possible schemes in Manchester and Birmingham did not occur.

Admittedley since privatisation there have been some new services Great Malvern - Brighton but I not sure if they are actual real innovations or way of gettng the maximum use out of the stock. 

[Worcester_Passenger]

Admittedley since privatisation there have been some new services Great Malvern - Brighton but I not sure if they are actual real innovations or way of gettng the maximum use out of the stock. 

That one is indeed mostly about getting the use out of the stock.

One of the few real post-privatisation timetable re-casts was National Express's recast of the Midland Main Line south of Leicester. The old BR^▸ (British Rail(ways)) timetable was beyond erratic. It almost looked as if the driver rolled dice to decide which of the intermediate stations should be served. There was a lot of "two trains every 90 minutes" - which is so easy for the passengers to understand. The NX recast had regular half-hourly fasts and half-hourly semi-fasts (the latter serving Luton, Bedford, Wellingborough, Kettering and M Harboro'). The fasts overtook the semi-fasts at Leicester so as to give cross-platform interchange. Very tidy.

But ... (always a but) to make it really work you needed to change the track layout at Leicester to make the overtaking easier. But that didn't get done. And they put little 3-car 170s on the semi-fasts. Perhaps the passenger volumes had indeed been poor under the old timetable. But they weren't under the new one and the 170s were very quickly upgraded.

But now DafT has got involved, and the nice tidy timetable has gone out of the window. One of the semi-fasts has been diverted to Corby, so there's only an hourly link up to Leicester. Yes, there's now two trains an hour to Sheffield - but it's a 20/40 headway out of Sheffield so it's a bit of a mess.

I grew up south of the Thames, so I grew up on a diet of regular headway services that were cleverly timed to provide links at junctions (you can tell - "semi-fast" is a dead giveaway). And the Southern used to rebuild stations if they weren't in the right place for the timetable, so as to make the timetable work. Leicester would have had its trackwork sorted if it had been on the Southern.

Cheltenham is a fine example of a place where the Southern would have moved the station. You want to be able to do a cross-platform between a CrossCountry fast and a semi-fast (send that round via Worcester, please) for all the connections to work - so that needs four platforms instead of the poky pair they've got at the moment. Plenty of space north  of the present station...

The other fine example that I do remember is Ely, in the days when King's Lynn - London was only every two hours. But the trains crossed at Ely, and met the two-hourly Peterborough - Norwich services. Nowadays, connections at Ely are every hour - but they don't work, so overall travel times are probably slower.

Meanwhile, Virgin allegedly went through 85 iterations of working up a CrossCountry regular-headway pattern - and still didn't get a workable timetable. Where are the services to Portsmouth or to Liverpool (for starters), or in and out of Gloucester?

And on the CrossCountry main drag, you have the stupid situation that the Birmingham - Leeds - Edinburgh does the stops at Tamworth, Burton and Chesterfield while the short workings to Newcastle via Doncaster ignore these. The simple arithmetic of dwell times means that you'd really like to have the stops on the short workings, so as to even out the overall times from B'ham to York and N'cle. Southbound, the Birminghams depart Newcastle at 35 and 44 minutes past each hour.

Sorry - this turned into a longer rant than even I intended. But I don't think that the railways generally are very good at timetabling from the passengers' perspective (which was always a Southern strong point).  Correction - for "don't think" read "know".

[paul7575]

Someone once posted elsewhere that the XC^▸ (Cross Country Trains (franchise)) timetable is basically the 'first fix' for the national timetablers, (ie the 'traditional' XC routes on the main NE/SW and NW/SC axes, rather than the more recent additions).

The reason is that they invariably require paths crossing, joining or leaving many other main lines, so hypothetical questions such as 'why not run the Bournemouth service on the opposite half hour south of Basingstoke' might have repercussions in the Manchester area...

Paul 

[Andrew1939 from West Oxon]

Railway timetables are very difficult to devise and move. CLPG^» (Cotswold Line Promotion Group - about) has a timetable working group with members more knowledgeable about these matters than myself so I am just an observer of their discussions. We have found that the TOC^▸ (Train Operating Company) timetablers do listen to our comments and have often been able to incorporate CLPG suggestions. However much of the suggestions that have been accepted have been minor tweaking. So many suggestions for the Cotswold Line seem to be difficult to implement because of difficulties off the line. One matter that has been of concern is that the 20 mile redoubling of the line two years ago has not resulted in the full level of improvements in punctuality that everryone had hoped for. This seems to be because far too many trains pass at the ends of the central double track, i.e. at Evesham and Charlbury, when there would be more flexibility in coping with train delays if more trains passed somewhere around the middle of the central double track. The problem seems to wrap around the pathing of trains between Pad. & Oxf. where trains so often follow each other at at short intervals of only a few minutes. To enable trains to pass on the CL at a more ideal position would mean changing the timing of paths from PAD^» (Paddington (London) - next trains) and this brings even more complications for which there seem to be no solution without a wholesale recast of the nearly all Paddington departures. There are also difficulties at the Worcester end where the line constraints make it difficult to get better Worcester departure times. Further complications arise for Up trains that have long platform dwells (which can be useful as recovery time when a train has been previously delayed) but have to be kept on the CL because of congestion at Oxford station that would arise from a better CL running time. Further complications arise because rolling stock is now used more intensively so that changing the timing for one journey impacts on the particular rolling stock's next possibly completely different journey on another route. Then there are the consequent staffing problems . . . Difficulties seem to go on and on. These timetabling problems have got worse over the last 10-20 years as the number of train movements all over Britain have in general increased enormously (apart from Melksham as Grahame points out). With the FGW^▸ (First Great Western) franchise now in turmoil and the likelihood of not being resolved for possibly several years, no TOC would want to invest a great deal of staff time in wholesale timetable revision with no future security for its own operations so I think these timetabling problems and consequent traveller complaints are likely to continue for some time.

[Worcester_Passenger]

Railway timetables are very difficult to devise and move. CLPG has a timetable working group with members more knowledgeable about these matters than myself so I am just an observer of their discussions. We have found that the TOC^▸ (Train Operating Company) timetablers do listen to our comments and have often been able to incorporate CLPG suggestions. However much of the suggestions that have been accepted have been minor tweaking. So many suggestions for the Cotswold Line seem to be difficult to implement because of difficulties off the line. One matter that has been of concern is that the 20 mile redoubling of the line two years ago has not resulted in the full level of improvements in punctuality that everryone had hoped for. This seems to be because far too many trains pass at the ends of the central double track, i.e. at Evesham and Charlbury, when there would be more flexibility in coping with train delays if more trains passed somewhere around the middle of the central double track. The problem seems to wrap around the pathing of trains between Pad. & Oxf. where trains so often follow each other at at short intervals of only a few minutes. To enable trains to pass on the CL at a more ideal position would mean changing the timing of paths from PAD^» (Paddington (London) - next trains) and this brings even more complications for which there seem to be no solution without a wholesale recast of the nearly all Paddington departures. There are also difficulties at the Worcester end where the line constraints make it difficult to get better Worcester departure times. Further complications arise for Up trains that have long platform dwells (which can be useful as recovery time when a train has been previously delayed) but have to be kept on the CL because of congestion at Oxford station that would arise from a better CL running time. Further complications arise because rolling stock is now used more intensively so that changing the timing for one journey impacts on the particular rolling stock's next possibly completely different journey on another route. Then there are the consequent staffing problems . . . Difficulties seem to go on and on. These timetabling problems have got worse over the last 10-20 years as the number of train movements all over Britain have in general increased enormously (apart from Melksham as Grahame points out). With the FGW^▸ (First Great Western) franchise now in turmoil and the likelihood of not being resolved for possibly several years, no TOC would want to invest a great deal of staff time in wholesale timetable revision with no future security for its own operations so I think these timetabling problems and consequent traveller complaints are likely to continue for some time.

If I interpret you correctly, ^80M was spent on the re-doubling without thinking about the consequences.

[rogerpatenall]

Another bit of a rant. It annoyed me when NSE^▸ (Network South East) publicity featured prominently '6 trains an hour from Victoria to Bromley South'. Nothing further, so the assumption to the lay reader was a train every 10 minutes. In fact they left at XX 20, XX25, XX28, XX50,XX55 and XX58 or similar. Still a reasonable service but the advertising I found devious to put it mildly.

[Electric train]

Train timetabling starts with some basics.

Each class of traction Unit has an acceleration curve, there can different modes within each class which can be switchable.  Where this loco hauled the number of coaches or wagons will change the curve
Each type / class has a braking curve.  The number of passenger type coaches are they pad or brake block (vacuum or air) / wagons (loaded / unloaded.
These curves are laid on the route against the line speed for the class and type of train also gradients descending as well as ascending come into play especially for freight.
The stopping pattern for the service.

This all used to be mapped out manually, I can remember seeing the timetable planners office in the General Offices at Paddington in the late 70's they had literally a set of curved templates, pencils and a rubbers

[Phil]

they had literally a set of curved templates, pencils and a rubbers

Curved pencils! This only serves to make their achievements even more impressive

[Worcester_Passenger]

This all used to be mapped out manually, I can remember seeing the timetable planners office in the General Offices at Paddington in the late 70's they had literally a set of curved templates, pencils and a rubbers

No wonder that in those days they didn't change the timetable unless they had to.

[grahame]

Not showing the curves (which are relevant on double track lines), but very interesting to show you the complexities of timetabling ... have a look at these graphs which show single line timetabling.   Click at the links on the bottom of each of them to add in trains trat only run seasonally or on certain days of the week, and remember that on a single track following trains can't usually set off until the one in front has reached the next passing loop.  You can see in places how a single freight train can lead to a considerable break in pattern

http://www.2d53.co.uk/graph/GRSHAB78.htm

http://www.2d53.co.uk/graph/GRDJPI78.htm

[thetrout]

Now I know this station normally only gets roughly 2 trains per 2 hours, but some in between also. But as a followup from Worcester_Passenger's post I have another example which is Frome. It can be a real congestion spot when things go slightly wrong. or very wrong. Here are a few examples, but first a description of the line.

The single track branch comes off the GWML^▸ (Great Western Main Line) and converts to double track (There is a third track but this is a Freight Only line I understand). There is another spur off the branch which leads to the old S&D^▸ (Somerset and Dorset Joint Railway) Railway which is still in use for Quarry Trains to Whatley. The line then converts back to single track, through the station single track (There is a second platform but it is trackless) Also you can see where a third bay platform would've been where what is now Jewsons Outdoor Stock Yard. Line continues as single track right back onto the GWML.

Now that is a bit of a mess for starters. But there are a few times in the day where there are a few trains within 10 - 15 minutes of each other. The Branch is also used as a diversion to get slow trains (e.g. Freight from Torr Works Quarry or Network Rail Equipment Trains) out of the way of a Paddington Fast (e.g. TAU^» (Taunton - next trains) - RDG^‡(resolve) Non stop). Now lets not forget the Quarry trains to Whatley as well

Here are a few examples. I'll use the 18:49 FRO^» (Frome - next trains) - WEY and 19:06 FRO - BRI^» (Bristol Temple Meads - next trains) as an example. The 19:06 has around 10 minutes scheduled slack in Frome I believe to allow a HST^▸ (High Speed Train) to pass (You can certainly here one pass during the time the service is sat there). Now assuming the 18:49 is delayed, they generally don't hold it in the double track just outside the Westbury bound end (Off the Main Line). So they allow the service to come onto the platform. Now lets assume there is further dwell time of a few minutes as is often the case as that is an evening peak service from BRI and BTH^» (Bath Spa - next trains) (17:49 and 18:07 retrospectively). During all this the 19:06 is stuck on the GWML with nowhere to go blocking any "Fast" services that want to get past. There is nowhere else to put the thing! Invariably this causes delays to at least 2 other trains.

Another example is when the signaller decides to release a Quarry train from Whatley onto the GWML, There are occasions that trains that want to get into the Frome Branch have to wait for it to come out because it is too long to stop before the branch. This causes delays for Fast services going BOTH ways on the GMWL as you have the Freight crossing onto the Up Main and a Regional Service Blocking the Down Main.

There are also other occasions where the 14:41 towards BRI is late (IIRC^▸ (if I recall/remember/read correctly) that's a Gloucester Train) and the 14:59 to Warminster is on time. The problem is that also comes from Westbury and Terminates at FRO and works the 14:59 to Warminster. If they decide to allow that train in first rather than keep it in the loop, you have the 14:41 stuck out on the GWML again. I can see the slight logic in that now you only have 1 train thats late, just even more so now.

There are also further examples of when the 18:49 is late. If it is very late, they'll hold it to allow the Exeter St Davids train to pass. I cannot tell you the amount of times has happened. I'm assuming so said train (It's a semi fast HST) is not held up when the 18:49 stops at Bruton and Castle Cary... The only problem? I nearly always wanted to connect to the Exeter St Davids train at Castle Cary If this ever happened you just had to accept that this was going to be a guaranteed rail travel voucher as you would end up with over an hours delay, some occasions over 2

Although there isn't the problem of a train in the opposite direction, the 17:47 used to fall foul of this as well, however this was more recoverable as you could make the Exeter Train. *edit - correction to service pattern in italics. My bad. Sorry. TT *

[grahame]

Since 1967, Monday to Friday services from Frome to Westbury have gone up from 11 to 14, and expresses bypassing Frome towards Westbury from 8 (10 in summer) to 18, but I suspect the infrastructure is the same or less. So there's far more likely to be something in the way at times of late running.  And there are now through trains from places like Great Malvern which wasn't the case before.  I have noticed that delays "up north" sometimes result in a Worcester turnback or a service split at Bristol, with the Weymouth train leaving before the same service arrives from Great Malvern.  All shows the importance of a robust timetable with recovery time and not trying to do anything "silly".

There's a suggestion in the ITT^▸ (Invitation to Tender) of an hourly Westbury -> Weymouth and that would be even more trains, but fewer terminators; probably helpful.  Putting a loop through the station could help, but who would pay for the footbridge and lifts?

This sort of story acts as a very real warning for us for a TransWilts service ... and the need for it to be robust.  The 40 minute running time which is possible on regular station to station timings can be relaxed to 45 or 47 ... and I would be happy to see northbound services allowed (without penalty to the TOC^▸ (Train Operating Company) for this delay) to be held for up to 10 extra minutes on the single track to let late running Bristol services go through. Southbound to Westbury is more of an issue, as TransWilts needs to connect for quite heavy passenger flows for onwards services.  What fun 

[Not from Brighton]

So in essence, in order to change a single service, all conflicting services also have to be changed; thus, it's impossible to change one service without re-writing the entire UK^▸ (United Kingdom) timetable!
This reminds me of something I once read (on Wikipedia or some similarly reputable source) that the scrapyard just to the west of Birmingham Snow Hill has had a once weekly service to South Wales since the 19th century. Is there any real evidence for services that have remained unchanged for such a long time?